Ai-Media Technologies Limited (AIM) Earnings Call Transcript & Summary

Good morning, and welcome to Ai-Media Technologies First Half FY '24 results webinar for the period ending 31 December 2023. Presenting today is AIM's CEO, Tony Abrahams; and CFO, John Bird. Today's format will begin with the run-through of the results, followed by a Q&A session. [Operator Instructions] I'll now pass to Tony.

Thanks, Mel, and thanks, everyone, for joining. I'd like to begin by paying my respects to the traditional custodians of the land from which we gathered here in North Sydney, the Cammeraygal people of the Eora Nation, and pay my respect to their elders, past, present and emerging. I'm joined as well by my CFO, JB. And this has been a really exciting half. It's been a really exciting half on a number of fronts, both financial and technological. First up, I think, is really the development in AI that everyone is very familiar with in general life. And for the very first time this half, we have begun to see results from the AI engines that are actually superior to the legacy human workflow. So for the first time, we're able to engage with customers, not about any kind of trade-off between substituting a legacy human workflow for a technology solution, but to implement a technology solution that is scalable and better than the legacy human workflows. As a result of that, we saw almost half of our total revenue now being delivered with technology sales and almost 2/3 of the gross profit. And obviously, that has flown through to EBITDA and the cash balance. I'll skip to this slide next because what we've really seen in terms of the growth has been a focus on that technology revenue growth. That is the number that is really powering the scale and the profitability of the business. It is this revenue line that's now grown to be 48% of the total and has offset a 7% decline in our traditional legacy services business to return us to top line growth of 10%. That growth of the tech revenue is absolutely supported by our LEXI revenue, which has grown 54% compared to the prior corresponding period. And that has led to an increase in gross margins to 63%. Keep in mind, when we listed, our gross margins were 39%. So in 3 years, we've significantly improved that. And of course, we've moved from an EBITDA loss to an EBITDA profitable position and operating cash flow of $3.1 million for the half, up on $0.5 million last year and a very healthy cash balance of $11.7 million net of paying out all of the final payments to the EEG and ACS vendors. So we really are debt-free with that cash balance that's ready to take advantage of a lot of the growth opportunities that we're excitingly seeing with AI being where it is. So a quick update, a reminder on our unique ecosystem. What AI Media does is we deliver the best AI technologies embedded within our customers workflow. It's the ability to embed this technology seamlessly within our customer workflow, that is the source of our competitive advantage. That competitive advantage begins with our encrypted encoding technology, which literally plugs in to the audio source from our customers. This is the benefit of it. We are acquiring the audio at the earliest and cleanest possible format. We're taking that audio, and we're then delivering it up to our iCap cloud, where we host dozens of the latest AI engines. We allocate the best AI engine to that content based on what we know to be how those engines perform according to different types of content. And then importantly, those same encoders decode that caption screen and display those captions either on the TV or the device or to any other application that our customer wants. And this end-to-end one-stop solution is really what has helped us to power the transition of many of our customers as they take the leap from human to AI-delivered language services. Unpacking that a bit more. So those 3 elements of the competitive advantage, it begins with the encoder group, and those encoders can be either physical encoders which is the Encoder Pro, which literally plug in with a wire to the audio source from the customer or they can be virtual like our Alta product or cloud-based, like our Falcon product. So any way that a customer can get us the audio, we can take that. We can ingest it into our iCap system, and we can deliver LEXI services back to those customers. Importantly, we are agnostic as to which AI engine we use. That is important because it is horses for courses. There are different engines that perform better on different languages and on different types of content. Importantly, as I said, the AI has improved out of sight in the last 6 months, but also for languages other than English, and we've made huge strides this half in both French and Spanish in addition to English in a way that we're calling those languages out for the first time. The blue chip customers and new wins on the right, you'll notice on the right, a, that it's a good collection of global brands. What you'll also see is that there's a lot of sporting organization. And what we're seeing from the development of the AI algorithm is that as the technology improves week-by-week and month-by-month, more and more settings that previously couldn't be delivered with AI can now be delivered with AI. And that tipping point certainly happens with live sport in the last 12 months. And what we saw off the back of our very first sporting group win with Major League Soccer, which we announced last year was really the snowball effect then of the other sporting organizations like the NFL, like the NBA coming on board. And we're really seeing that, that snowball really continue to gain momentum. And then with TV Azteca and Telemundo, they're very large Spanish broadcasters that we've switched across to LEXI and are now using LEXI very happily. Al Jazeera has moved across from being a services customer to being a technology customer. So all of that revenue has now gone from services and is in our technology bucket. And then the other 2 to call out is the House of Commons in Canada, which is a bilingual English and French LEXI setup and MultiChoice, which is our first South African broadcast customer. So lots of wins across the board. If you recall, our strategy was to extend the dominant position that iCap had in U.S. broadcast and take it out of the broadcast industry in the United States and then take it global. And I think we've demonstrated very clearly the appetite of customers in countries outside the U.S. and in industries outside of broadcast for our LEXI service. And we're seeing that with sustained growth in the iCap network. So every time a new customer connects to the network, we get an extra point on the iCap network and that continues to grow. And of course, the percentage of share that LEXI is taking of that iCap network is growing even stronger. This, for me, is probably the money slide of everything. This is actually where -- this is the proof point of the data that demonstrates very, very clearly that the AI is continuing to advance and shows that in the last 6 months, LEXI 3.0 has, for the first time, on average, outperformed our human captioning. And what we're seeing is that, that curve continued to move to the right. And what we're therefore seeing is through our LEXI Toolkit more and more opportunities for AI Media to introduce new AI-powered language solutions for our customers off the back of what has been a very, very successful launch of LEXI 3.0 in the live environment. In terms of the operational update, the growth strategy is really fourfold. The first is to continue to grow LEXI, and that involves both transitioning existing customers like Al Jazeera, but it also obviously involves signing on new customers like MultiChoice and the House of Commons in Canada. We are looking to grow that iCap network globally, and that's both outside of broadcast and outside of the United States, and we're having success there. On the bottom right, it's about continuing to extend and enhance the LEXI Toolkit. And what this is about is about bringing together many more applications that AI can help language services above and beyond just live captioning. Live captioning is probably the hardest of the disciplines to actually get right. And once that has been done, what we're finding is that the same competitive advantage that sees us getting the best possible quality audio into the system, matching it with the best engine and delivering it back seamlessly embedded into customer workflows, we're seeing there's also a market for that for recorded now. And recorded captioning is an industry, a market that's many, many times the size of the live environment. When you think about it, most people consume media in a non-live environment and having a way to deliver a caption file back to a customer with a single click very, very fast. So what we can do with LEXI Recorded, if a customer has got a 30-second clip, we will get them it captioned clip back within a minute. There's no way that you can get that done with any kind of human in the loop. And we're also finding that the quality on that turnaround is excellent because of all of the investments that we've made in LEXI Live, that's also coming through in terms of LEXI Recorded. So that product launched this month. And we've already got over 30 customers signed up and using that. It's small fry at the moment. We're not expecting this to have a major impact this half in terms of revenue, but it is a really, really important growth lever for the business into FY '25 and beyond. What we are expecting to have a more significant lift this half is our brand-new LEXI Disaster Recovery product. And this is the one that I'm most excited about. I think it will deliver the fastest sales growth of any product that AI Media has ever launched. What it is, is it's a response to effectively a request from customers to completely switch off their human captioning legacy workflows. And a large part of why they're maintaining their human workflows is in the event of disaster recovery. Why is this important? Well, the iCap infrastructure is a cloud-based infrastructure. And the cloud-based infrastructures are reliant on the Internet. And so when you talk about very, very large events like the Super Bowl or the Olympics where you add enormous amounts of strain onto the Internet because you have hundreds of millions of people or billions of people streaming video at the same time, we need a solution that isn't reliant on the Internet in order to give our customers the level of comfort that at the moment, they get around by having dial-up modem for captioners who are sitting at home. So we're talking about replacing a dial-up modem backup solution for customers with an on-prem piece of hardware that in the event that the Internet goes down, will either automatically failover or allow for a manual red button pressing case of emergency button. We are shipping that product today out of our headquarters in New York, and we will be officially launching that product at the National Association of Broadcasters trade show in Las Vegas in 6 weeks. This will allow us for the very first time to offer a true end-to-end technology solution with automated failover that will allow broadcast customers for the very first time to switch off those legacy workflows. We have considerable back orders on this product, and I think it's fair to say we are expecting to garner a lot of interest once this product is officially launched in April. And then the final piece of our operational growth strategy is to continue to grow our indirect sales channels. What we are finding is that particularly with our encoder sales, those encoder sales are very, very much skewed towards when a broadcaster is upgrading their broadcast infrastructure. And when they upgrade that broadcast infrastructure, they typically employ a systems integrator to help them find the best pieces of kit to put together. And so having relationships with those systems integrators is key to ensuring that we can unlock all of those opportunities as well as being quite critical for those new sales opportunities in markets outside of broadcast and in territories outside of the United States. So finding the expert in Eastern European broadcast in countries in Asia, is a much easier sell than trying to establish a direct sales team there. So we're looking at both growing the direct sales team, and certainly, that's been a component to increase costs in the half as well as, obviously, the work that's gone in for launching LEXI Recorded and LEXI DR. But we're also focused on growing those channel partnerships and obviously paying out commissions in return for sales for those partners as well. Then what this diagram really shows is how we're looking to branch out from LEXI Live, right? The success that we've had in LEXI Live is really just talking about that live captioning market. We've launched that product. It's going very well. We're now at the point where it is on the tipping point of taking over from a lot of the legacy human workflows because it's cheaper, faster and better. But then what? And what we're seeing is actually that this leading edge of AI technology and being at the forefront of this AI technology really puts us in the box seat to work with our customers on additional products that they may want to integrate within their particular workflows. Whether that's translation, disaster recovery, LEXI Local is really key for our government customers, and we've just won a fourth high-security U.S. government department to deploy this LEXI Local solution on, which is fully secured behind their firewalls. We don't see any of the data. And then, of course, LEXI Recorded and LEXI Library, which gives our customers access by searching any word to their library archive. And all of these things together, we're looking to continue to grow and to grow out the features as the AI gets better and better. So in terms of our 2024 calendar year product road map goals, it's on the encoding side, making sure that we have full global functionality of those encoders and that really is market by market, country by country, doing the painstaking work. We are certainly there in the U.K., Australia, New Zealand, France, Germany, and we're getting very close on a number of other Eastern European countries, and we are compliant, for example, with the Philippines now. So it's a question of us going methodically country by country, to make sure that we have the language set employed in the decoding and encoding technology, but also that the captions appear in the correct way according to the local broadcasting standard. What we're also looking to do is enable text to speech across the iCap network so we can begin to introduce services that focus on voice AI. So these are services like automated dubbing services and audio description. We're also working on new versions of our hardware that don't need to be plugged into a broadcasting rack, and that could just sit on someone's desk or be carried in their handbag. In terms of iCap enhancements, a lot of it is related to security and then third-party integration. What we have done now with iCap and many of you will remember that when we acquired EEG, there was no hourly charge in the market for iCap. And so a lot of our competitors that were using iCap were actually free loading off our iCap infrastructure. We've now successfully employed charging per hour for iCap and all of our competitors have begun to be invoiced for that service effective from the 1st of November. It's not a huge amount of revenue for us, but it certainly does give us enough to continue to invest in that really critical elements of our competitive moat, but it also demonstrates the value of iCap because our competitors are actually now paying for it for the first time rather than getting it for free or putting something of their own in. I think if any of them had a competing product to iCap that would have worked, they certainly would have preferred to have implemented it rather than have paid off money, and I do want to put a particular shout-out -- give a particular shout-out to JB and the team who did a great job on those negotiations as we went through what was a tricky process of charging for something that was previously free. And then the final product road map goals are really focusing on that LEXI Toolkit enhancement. And you can see it's the LEXI Disaster Recovery, which is the fully redundant captioning solution and an opportunity to sell additional hardware to existing customers. LEXI Recorded, the ultrafast recorded media captioning solution. Further work on LEXI Translate, so watch this space. And then using AI in ways to continue to improve the outcomes of LEXI, which is focusing on the things that at the moment, it doesn't do well but we know that it can do well in the future. And that's like using the generative AI to help with topic modeling, to automatically categorize and recognize music, background sounds, also, automatic language detection, we're actually making significant progress on that already, and then ultimately, audio description and dubbing solutions. So utilizing not just voice to text but text to voice and making sure that, that can get carried across the network. So building out effectively a full suite of AI solutions off the back of the success that we've had in live captioning. And I will hand over to JB for the financials, JB?

Yes. Thanks, Tony, and thanks, everybody, for turning up this morning to this. I think this is more than anything indicative of where we said we'd get to, and I'm pretty excited about the fact that we've actually achieved what we set out to do in this half. We continue to see a slight decline in services revenue. And that will continue. We'll transition Channel 7 and Channel 9 to be -- using LEXI. And those themselves are probably about 15% of that revenue. So we'll continue to see that decline. But as it moves, it's gaining that higher-margin tech return. So we're sort of getting around the 80% mark on the tech business, and that's where we want the entire business to move to. And I think that's what we're about. We bought a technology company. We continue to invest in technology, and that's now best evidence is growth in the technology revenue, but more importantly, we're actually growing the overall revenue. As you get a bigger base in the growth engine, we're seeing a return of -- whilst 10% is good, we had to continue to drive that revenue even further. And that, of course, flows through the margin. Our OpEx, it is up. It's up about $1.7 million. We did have a couple of 3 kicks. Last year, we had a gain of about $400,000 in our FX, which just goes into the OpEx realized gain. So the bulk of the increase is probably associated with people. We still have a reasonably large staff around the world, and those people, particularly in the engineering space, we're getting last year in excess of 5% increases in the engineering space. We've also grown the number of engineers. As Tony said, we've invested in product enhancement. So rather than capitalizing these, we treat them as an expense. I'm a bit of an old-school finance guy, unless there's an absolute correlation, I'd rather not capitalize it. I'd rather take it through the books. So we're pretty conservative in that space. And the benefit of that is that you see the revenue grow, and we will just have a static expenditure line. We've also invested since the comparative period, 12 months ago, we've also invested in about a dozen salespeople around the world. One of the interesting things that Tony mentioned is we've got a large number of sporting organizations, particularly in the U.S. That investment is what's paid those dividends. Those people started 18 months ago -- 12 months ago and it takes that long to move some of these larger organizations, like the football league over in the U.S. across to buy LEXI. So now we're seeing the fruits of that investment in the top line. And I think that's actually probably one of the most exciting things is the sales growth in the technology space just continues to do us proud. It's where we've invested, and it's where I think we'll continue to see the dividends. I think on this slide, one of the key shout-outs of course, and I'm happy to go through the waterfall on to the next slide is our cash. We continue to sustain a good cash return as we've improved the business. I'll explain some of those things there. But this is after we've paid out the final payments for the acquisition, particularly of EEG, but AUD 8.1 million went out the door for that. No regrets there because it's playing big [differences]. Tony, the next one. This is simply -- it's very evident what we're referring to there. The increasing split in the revenue and the dominance, if you like, of the technology sales, that reflects very well in the gross profit. So as we continue to see the development of the technology business, we're seeing an increasing return for the business. And I think as said in some of the other slides, we're getting early 40% for the services business and around the 80% of the technology, which is -- the services business is still a magical number based on the fact that we had lower than that for -- in the original IPO. So both sides of the business performed well. But where we're growing is the one that's the best business, the technology business. And of course, one of the things I'm very proud of is the actual increase in the cash balance. We had a -- I joined AIM 3 years ago, and we had a number of issues in the way we were managing the business and our receivables were extraordinarily high. We've now brought those into line with what I consider as best practice. And that, in fact, is probably the biggest non-EBITDA increase in cash is the ongoing attention to trade receivables. As you grow the business, you do see a little bit of cash go out in inventory. We have invested even in the half in the disaster recovery box. And we've also built up enough inventory to start selling it. As Tony mentioned, that piece of equipment has started going out as we speak. And that's really exciting because that's our next big, I'll call it, a mass sales box, and we expect that to do pretty well in the coming years. Thanks, Tony.

Thanks, JB. And in terms of then FY '24 growth priorities and outlook, again, this is coming back to those same 4 slides that we're talking about. We're going to report on these key metrics and key drivers of success. So in terms of growing LEXI, it's threefold, we're completing the transition of our services customers to LEXI. Al Jazeera is one that we did last half, 7 and 9 will be looking to do mostly this half. Win new customers in those established broadcast markets and we continue to do that with customers like TV5. In Quebec, which was our very first LEXI French installation. Off the back of that, we had the guy at TV 5, say, "Oh my God, LEXI is faster and better than any human, I've ever had to do any captioning, and I'm saving $500,000 a year." Well, that got TV5 in France interested. And that's how -- that's the domino effect, right? That's the snowball effect of first getting the first customer and then having an AI product that can seamlessly embed in customers workflow without them having to invest millions of dollars to upend their existing workflows and then making sure that, that is future-proofed, finding an index customer who can then trial it for 3 months and then use them to go and help us to sail 2, 3, 4 and 5 in that particular market. That model is working really well. You've seen it with the sports broadcasters, you've seen it with the French and Spanish language and now we're looking at doing that in those other markets. And then, of course, it's introducing LEXI to the enterprise customers, who we also see as having a big need for this. And probably the biggest one to call out in terms of we haven't yet got an index customer but we should do by the next set of results, is to bring to churches, is the religious organizations, all of whom are looking to implement this solution. We were just at the National Religious Broadcasters Association Conference in the States the week before last. Huge interest in a solution that's actually going to help them deliver a service that's going to increase the engagement of their congregants but also one that's going to allow them to translate this message particularly for a lot of the Chinese and Korean churches that we're speaking with over there. So there's lots of opportunity outside of what people would think of as traditional broadcast. And we've always said that, and we're starting to see those results. So really, we ultimately see these markets of broadcast, government and enterprise as having equal scope for us to grow. And so we're very much beyond broadcast, but we're starting with broadcast because effectively, that's the easier sale because it's a substitution sale because they're already using and paying for captioning, and this is a no-brainer because this allows them to save money. But beyond that, certainly, we are seeing parliaments, courtrooms, legal, that kind of thing, very, very interested in this kind of solution, albeit none of that is going to really hit in FY '24, but we're really setting the business up for success in the years ahead. That involves growing the iCap network globally, new territories, new industries and improving the scalability and resilience of the network while maintaining its security posture. Extending and enhancing the LEXI Toolkits, I spoke about Lexi DR and Lexi Recorded. What we are also going to be launching in this next financial year in September of 2024 is a LEXI Audio Description product. So a product that previously would have required 20-plus hours of human intervention, we are looking to deliver an AI solution that is fully automated. What that does is it creates an extra audio track so that someone who is blind or has low vision can understand the visual elements of the storyline that otherwise they would miss. So if you think about it, it has to be done in a gap in the dialogue of the particular show. And it has to figure out what is the salient information to transcribe and speak. So for example, Clark Kent goes into a phone booth, spins around, comes out as Superman, grabs lowest and flies off. None of that is spoken, but all of it would need to be described during that gap in the dialogue so that someone who is blind or low vision can follow the story line. We are looking to automate that process with generative AI. So you can look at the generative AI tools that we're seeing now through AI video and all of the components that we need to pull together are in our LEXI lab at the moment. So we're very excited to be launching that. We think that will be a game-changer in September of this year. And then as I said earlier, continuing to grow those channel partnerships with a focus on the systems integrator relationship. And this is all about the technology revenue. This isn't about any more growing the services revenue. We do expect to maintain that services wrapper, if you like, around the technology business for the foreseeable future. There will always be small amounts of content that you do need that human wrapper around, but that will be an ever-declining share of the total, and we're expecting that, obviously, that tech revenue will grow very much more strongly. And then just to cover off a little on the competitor table and then I think we'll be in a position to take Q&A. So there's been quite a bit of public information about Red Bee, 3Play and Verbit on that screen who are our traditional, if you like, services competitors. These are 3 of the larger companies that we have migrated across to iCap charging. So we are -- so we know for sure that they don't have a competitor iCap product. It's something we didn't know for sure this time last year. We know that they weren't able to build one. We know that they are paying us now for the privilege of using that. But we also know through our public disclosures that their revenues are going backwards. A part of what they're saying their revenues are going backwards is because of the AI tools, in particular, in the live space. And that's just -- we're seeing that on the other side of the equation, and that's what's helping to propel that 38% growth in our own technology revenue is obviously -- yes, we're cannibalizing some of our own services revenue, but we're cannibalizing far more of our competitors services revenue as we transition across. And that's it, I might now hand over to you for Q&A.

We have a few questions that have come in. So firstly, in terms of the language services the other language services, French, Spanish and English extend to translation, i.e., English to French?

So that is the LEXI Translate product. What I was talking about today was actually same language transcription. So we have, for a long time, had pretty good traction with our automated translation engine going from English into French. It's actually a much bigger market going from French into French and from Spanish into Spanish than it is to do the translation because most people when they're watching French TV, watch it in French. And most people when they're watching Spanish TV, watch it in Spanish. And we've never before had the level of success with LEXI that we have seen with live in French and Spanish. We've already had good translation. So the breakthrough is actually in the live to-live same language space.

Okay. And just following on from that, Ken had a question in regards to languages in terms of, does this extend to other languages using different characters, i.e., Chinese?

Great question, Ken. Like great question. That's really hit the nail on the head. Like I mentioned before, that we were working as a product priority this year to ensure that the encoding technology worked as well outside of the United States as it does inside the United States. An element of that is ensuring that we have all of the character sets from all of those different languages embedded within the iCap code. And that is something that needs to be done language by language, country by country. As well as then working out how those character sets are then displayed on the TV set or on your mobile device. And in particular, if you think about Chinese, you need to make sure that, that Chinese character scales depending on whether you're watching it on a mobile or you're watching it on TV. In particular, if those characters are too small, they become illegible very, very quickly. So the dynamic resizing and making sure that you've got the right number of those in the right place on the right line, that's all part of that encoding technology. That's a great question.

And actually, we have a question from Nick Harris, so I'll just allow him to ask that directly.

I've got a couple of questions. And congratulations on that top line revenue growth and free cash flow growth, they are great. I might just shoot them off. So LEXI 3.0, I mean, that's a bit of a game-changer from what I can see on any hour, faster and better than humans, if I'm reading that right.

Yes, you're reading that right.

Wow. So it's obviously super new, but it sounds like you've kind of soft launched it with customers and I'm just trying to understand that a little bit more. So just so it's crystal clear to me that LEXI DR where you've now got the tech being better than people. Is that kind of this missing length that will now allow businesses or end customers to sort of switch off the legacy stuff and switch on to captioning with a safety blanket that you've still got redundancy? Is that kind of a key point?

Yes, you've nailed it. So firstly, I just want to clarify, LEXI 3.0, we already launched because you mentioned at some point that, that was a soft launch. We actually launched LEXI 3.0 in April last year. And then -- but what we've got now is we've basically got 9 months of data on this being used in the field. So that's what's enabled us to -- and it's improved over that 9-month period. And we've improved our implementation of it. So we've got better isolated audio, we're looking at better places to find the topic model for LEXI and so on. But that's what's seen in that graph continue to move over to the right. And yes, that's exactly right about LEXI DR. So at the moment, customers can't switch off all of their legacy infrastructure in case the Internet goes down, and the internet goes down all the time, like there's no way to control for that. So the only solution to allow -- and so the existing solution for a customer is to get someone to use an old-fashioned modem to connect via a phone line to the encoder and therefore, when the Internet goes down, that's the current redundancy. Now that's very expensive. You have to have someone on standby. You have to maintain all the modems, you have to maintain all this legacy infrastructure. So exactly, as you said, Nick, LEXI DR is a safety blanket, which will allow our customers to decommission all of the legacy modem infrastructure, and reliably move across to an all-technology solution, and that will itself release significant cost savings for customers because they will no longer have to maintain effectively a human legacy system just for backup. And that's why the customers are so excited about it. That's why we've put so much cost and effort in the last half to getting this ready to ship. And it ships today. So the very first customers that are getting the soft launch of it are getting it today. But the official launch will be in 6 weeks at the National Association of Broadcasters show in Las Vegas. Does that make sense, Nick? Any follow-up?

Yes. Yes, that's awesome. And so I mean, logically, not to adjust anything, but just understanding the logic, that should really fast-track the adoption of your technology solution because that's been the kind of bottleneck, I guess, the need to have a plan B?

Well, no. I mean, first customers will use both A and B because they'll just want to dip their toe in the water to it -- because I mean, they -- customers pay on the basis of CapEx and OpEx. There's both CapEx and OpEx related to maintaining kind of the backup infrastructure. So you need the equipment that's going to be able to take a modem call, for example, right? You need to maintain phone lines that are going to take a modem call. All of these things are getting increasingly expensive because they're kind of yesterday's technology. It's like those banking programmers that are the only ones that can fix the 1970s mainframe at the CBA. They get paid hundreds -- they get paid over $1 million a year, sometimes to get the kind of -- because they've got this legacy system that needs to be kept up to speed. Now that's a risk, right, maintaining modem for customers. I mean, that sort of turns off as a major risk for them. So being able to completely switch off the solution once we've proven that LEXI 3.0 is better, faster and cheaper, that's the point at which customers then say, "Okay, can I have the safety blanket as well? Can I turn off the modems? Can I turn off the dark fiber, et cetera, et cetera." And we're just getting to that point now with those customers that have got 6-plus months of LEXI 3.0 experience. So we're expecting that to continue and to snowball. And then yes, this would be part of then the integrated solution for those customers that are kind of leapfrogging the kind of mixed mode and they want to just jump straight to technology, then this will, yes, be absolutely part of the bill of materials for that change that we'll sell through to the systems integrators largely. JB, did you have anything you wanted to add on to that? I know you spent a lot of time over in the...

No, I think that's answered it, Tony. And I think that can be dovetailed into some of the other questions, too.

Nick, you had a -- that was part...

Yes, that's great. So I mean, logically, it's opening up a bigger market as I see it. That's the bit I wanted to clarify. It will take time, but the barriers to adoption become a lot lower, so the potential for total addressable market should be a fair bit bigger. Is that a fair...

I mean, that's even just a subset of our overall strategy with the LEXI Toolkit, which is fundamentally all about growing our addressable market within language services. So we're starting at the pointy end of the pyramid with live captioning and then we're pushing down as we go to the other elements, like LEXI Recorded, like LEXI DR, like LEXI Translate, like LEXI Audio Description, like LEXI Library, right? It's about using this fundamental position of strength that we have with the encoders and with iCap to now effectively sell more AI tools to our broadcasters who are increasingly going to benefit from the introduction of them within their existing workflows. And that's our special sauce. We can deliver the latest AI technologies quickly to our customers without them having loads of OpEx because of the flexibility of our implementation within their workflow.

That is great. And just one for JB. Congratulations on the free cash flow. Obviously, you've really pulled in your payment terms, which is great. So I just want to clarify, you've obviously worked on that for 3 years. I presume you're most of the way there now. We shouldn't expect too much more working capital gains moving forward. It's kind of where it should be. Is that how we think about it?

That's certainly how I think about it, Nick. We had to standardize across 5 acquisitions. We had to do customer onboarding. We had to do systems changes. All of those investments have helped us get us to this point. And I think we should be able to hold it at this point. So there was a question about will that reverse? No, it shouldn't reverse. But I don't think that we will go from 45 days down to 35 days on the debtors basis, mainly because we actually offer a 30-day terms. So I think it's a credit to the entire organization pulled together, and we've managed to get everybody from -- right through the organization, including our CEO has rung up to customers to help us improve that process. And so I think that's a really good outcome, and I don't expect it to change.

Well, it's not a sale until you collect the cash, right, JB?

Correct. Correct.

Thanks, Nick. And then now maybe moving on from that line of questioning. We have a question here that says, impressive results with LEXI 3.0 accuracy being better than humans. Are your competitors seeing similar results with their automated captioning offerings?

A very good question. Very, very good question. So I will pivot to a competitor that is probably the most similar to us in an environment which is Microsoft Teams. So Microsoft Teams is not iCap. So our -- I mentioned before that Microsoft are a key customer. It's not across then Microsoft Teams product although we do integrate with Microsoft Teams. Fundamentally, what we have seen in meetings with Microsoft team is the AI captioning explode and can almost completely replace human captioning. And it's interesting that the analogy with the Microsoft Teams setup is very, very similar to how we've architectured iCap, and iCap is designed to work outside of Microsoft Teams. But it is probably the best direct competitor, if you like, that has absolutely seen the take-up of the automated tools. I think they are doing billions of minutes a month on automated captioning is the latest number that I've had from Microsoft Teams. And it has almost completely displaced humans, both because it's very low cost, it's immediate, but it's also on demand. So someone can start a meeting in 5 minutes' time and have the captions available at a click. If you have to book a human, that's never going to be possible. And so a lot of it just is about that flexibility and that on-demand nature and that integrated nature of it. What's the similarity? What's the key similarity with Microsoft Teams? Well, they've got those 3 elements -- those same 3 elements that are present with iCap. They've got the best possible encoding. They're taking the microphone directly from each meeting participants. They're using the best AI engine, which they've got available to them. And they've got a whole bunch of metadata associated with you because you're a Microsoft customer. It knows who all your colleagues are, it's read all your emails, it understands the context of the meeting that's coming up. And in a very similar way, that's how we actually architect the LEXI Data Exchange services. So that gives further paid for the kind of technology migration point. And we're looking to effectively deliver that same level of migration outside of the team's environment.

Thanks, Tony. And then we have a few questions here from Gabriel. So I might just start with the first couple that start with strategy. Going back and listening to some of the interviews you've done, it seems you haven't pursued the consolidation that some others have, choosing instead to stay in a broadcasting niche. Is that your attitude as to the overall strategy vis-a-vis grow? And then...

Sorry, Mel, I have to correct something there. So firstly, we did do 4 acquisitions. So it's not like we haven't been acquisitive, but those 4 acquisitions have helped us get to the place where we are now. And in particular, the EEG acquisition has given us all of the final puzzle for delivering a technology solution according to the next generation of AI technology, not just in broadcast. I can't -- I'm trying to stress that, it's not about broadcast. It's about live. It's absolutely about live. But this technology will work outside of it. It does work outside of broadcast. We're using it in education settings. We're using it in legal settings. We're using it on cruise liners. We're using it on airlines. We're using it in government department. This is about anywhere that people want live transcription, translation and captioning. It's got nothing to do with the broadcasting niche. So I just want to clear that up first.

So what we might do is we might move to Gabriel's questions around sales, marketing and customer acquisitions. Are the discussions with customers around the usefulness of captioning and the product generally? Or are the discussions with customers around whether it's worthwhile paying for accuracy? So what I'm essentially shooting at is that you're interacting with a sympathetic audience and so the discussions around competitive pricing? Or is your experience that you are having to sell and convince people 2 things: the notion of captioning as a value-added feature, and namely your product as a solution?

Good question -- sorry, JB, you want to take it?

Do you want to break and I'll answer one because ...

Okay. Go for it.

I mean -- it is a good question because this also dovetails back to what Tony was saying previously about broadcasting. And the depth of what we do comes about because of 2 things: in the broadcast market, highly regulated, and this will go on to another question that's answered, you have to get above certain NERs or scores. And it's highly regulated right around the world. So there is a need for these companies in that sector, which we're calling broadcast to actually have captioning. Us being so much cheaper than humans and so much more accurate will ensure our dominant position. The depth of language, i.e., French as just one example means that we've got further access points. So all of those things are helping that. Then there is -- the whole world has moved through to inclusion. And inclusion is important in every facet of, firstly, work, but generally life. And so you can't have an arena or a church or a parliament that doesn't give access to everybody. And so this is one of those things that people are knocking on our door saying, how do we get this, what's the capital cost, what's the operating cost model? And compared to alternatives, the particularly used humans, they now understand how cost effective this is.

And Tony, did you have anything else you wanted to add there?

Yes. I mean, just to summarize, I think the question was, are we having to convince customers of the benefits of captioning? No, we're not. They all get it. Are we having to convince them of the possibility of AI? No, they all get that. That was the big change from last year. Now they get that AI really can be as transformational as we've been saying since 2021. Now the customers get it.

And then next, it's for the encoders in particular that you're selling to the streaming services, can you talk to me about the billing? Is it a LEXI suite as an ongoing charge with the cost of the encoder bundled in? Or do encoders have their own separate charge and the captioning as a [potential] license with recurring charges for the service?

It depends on the customer and part of the flexibility is making sure that we package up a solution for them that works for them. So if they prefer CapEx, we can CapEx. If they prefer OpEx, we can OpEx. But most of the benefit of signing up a new customer is not in the encoding revenue. It's in the future stream of LEXI revenue. And so in a sense, what we just need to do is we need to get more customers signed up more quickly through the encoder network so that we can turn on more LEXI more quickly.

And the final set of questions from Gabriel is TV Intuition, is that broadcast captioning has a kind of embedded demising rate of return that a viewer can recognize the difference when what's appeared as a caption isn't what spoken? And so audiences [indiscernible] to 99.8 plus versus 99.9 accuracy, for example, hailing in [indiscernible] right and what's your expectations around the incremental gains in accuracy and the competition for low bids for poorer quality services?

Yes. I mean, that's a great question, right? And that talks -- so firstly, is there a huge difference between, say, 99.9 and 99.8 on a live TV program? No, there isn't. So is there additional benefit in getting from 99.8 to 99.9? Not much, right, because you're not actually going to convince anybody else to do it. Of course, it does help because it improves the overall engine and the better the overall engine is, the more likely you are to turn more customers on. And the more likely it's going to perform on or better on really, really difficult content. But what happens when you've got the live market delivering those kind of accuracies, is you then start to turn on the power for the recorded media market, which those kind of levels of accuracy are what people expect for recorded. They don't expect it for live. But that recorded market is 100x the size of the live market. So that's why we've launched LEXI Recorded now. We want a slice of that much bigger market and the place we're chasing first is fast. So where people need things turned around fast, they're prepared for a lower quality. If we can get to that level of quality that you're talking about, three 9s, then that's all recorded media market opens up to us, and that's where we want to be in a good position when that tech is ready. Is the tech ready for that yet? No. Are we talking a small handful of years before it probably is? Yes. And that's why we're expanding the LEXI Toolkit, not all about live captioning, it's well beyond that. It's about AI language services for broadcast, government and enterprise customers right around the world.

Our next question comes from Stella and the first one is technology related. Are there any technology advancements that threatens EEG's encoders important? As audio and video over IP ecosystems grow, integrated networks grow and now increasingly move audio video onto cloud directly. Meanwhile, analog to network encoder decode has since declined in relevance, how does EEG encoding technology fit in this landscape?

Well, firstly, there's nothing analog about our EEG technology, right? So there's -- none of the technology is analog. The serial digital interface is most backward leaning technology and that's still the largest by number implementation of broadcast systems around the world. Then what we've got is we've got encoding technology for all of the new and upcoming video standards, including IP 2110 transport stream and RCMP streaming or just basic web streaming over the Internet. When you say, is there something that's going to threaten the EEG coding devices, I just want to be clear that where there is a walled garden that already exists like Microsoft Teams, you don't need an encoder because the audio is already getting piped into that system by virtue of the overall ecosystem. What we do is we provide that encoding technology outside of walled garden, and there's plenty of circumstances like that, right? So that's the systems integrator. That's where you're putting bunches of different kits into a new build, and you need to make sure that it all seamlessly integrates with each other. Sure, there are competitor solutions that do that, ENCO and Ava are some of those that we've demonstrated. But we are being much more successful than those players in the markets that we're targeting at the moment. And we're not seeing any kind of competitive threat gaining traction in the markets in which we're operating. There are competitors, they do offer similar services, but our products are proving better and better positioned in market, notwithstanding they're a higher price.

And then moving on, we have a question about how we think about baseline services revenue. JB mentioned 15% more services to transition, so maybe another $5 million out based on the half run rate. Should we think about that at $20 million or $25 million as a mothballed business unit?

It will continue to decline, right? And the important thing around that is that we'll continue to cut the cost so that it continues to make a contribution to the overall business going forward. We're not making any forecast as to how quickly that's going to decline or how long that's going to decline. We are focused on growing the technology revenue and investing in that. Yes, we will maintain that services wrapper. Yes, there will always be a services wrapper. Exactly what that number is, whether it's 20, 25, 15, I can't tell you.

Just to add, one of the transitions that I was mentioning was moving Channel 7 and Channel 9 as examples from services to fully LEXI. And that turns into a technology customer. So that will be part of why there's this attrition of services revenue that Tony mentioned.

Next question from Julie is, will LEXI 3.0 capability offer opportunities for captioning cinema?

That's a great question, Julie. Thank you. So it's interesting because the issue with -- so LEXI 3.0 firstly is just live. But then the question is, I guess, would the improvement in LEXI 3.0 leads the ability for us to put captions on movie content in cinemas? Actually, that's not the problem. The problem in movie theaters is movie theaters don't want to put words on screen when half the people in the movie theater don't want to see them, and it might go over key elements of the dialogue. Those caption files already exist. And in fact, some movie theaters put on open caption screening, and that's what we're seeing with movie theaters. Is there an opportunity to perhaps stream captions to individual viewers that are sitting in movie theaters? That's been tried with a system called CaptiView and a bunch of other things, and it gives people headaches because they have to watch -- switch their focus from being like on their phone or on their device to on the screen to back to their device and it turns out that gives you a headache, which is not what people want when they go to the movie theater. So I think in a movie theater, what we need is we just need those movie theaters to show more accessible screening. But of course, the window for release on cinema is now getting shorter and shorter and shorter to the point where it's going to be on Netflix or Stan. So a lot of people who need captions just wait the extra 1.5 months until it's available on their home cinemas and then they turn the captions on. But it is an example of -- it's a great question to illustrate just how all that last mile really matters in terms of actually getting this content with captions in front of you if you need it.

And JB, we've got a question here from Francis. What was the context behind the decrease in trade receivables in the half that had a positive $2.4 million impact to cash? Should we expect this to be normalized in the next half year? For example, a negative circa $2 million working capital impact?

Yes. Look, I think the second part of that is the critical question and the answer to that is, no. We have really tightened up the processes and the $2.4 million is -- it's effectively a one-off hit. If we keep increasing our sales, then working capital will change a little bit. But I think for me, the critical element of that debt is that we bought the debtors days down from 79 days at one stage -- just under 80 down to the low 40s, and that's partly why we've had this benefit this year. So I don't expect that to be wiped out, if you like. I expect us to just continue with a very tight approach to debtors.

Our next question is, congrats on success of LEXI over race data, accuracy and traction. What impact does this have on cost of sales and importantly, operating expenses with first half being $18.6 million? What is the scope, the cost out as rate state of LEXI reduces as part of AIM? Is it a material number? And if so, when would we expect to see the cost out results?

Yes. Look, I might answer that one, too, Tony, if that's all right. There's 2 parts of the cost argument, of course, as we see the decline in services revenue. There is the direct cost, which is cost of sales. A lot of that is a casual workforce or crowd workforce because a lot of them are crowd, but there -- we do have a number of permanents, particularly stenographers, which is a relatively highly skilled but relatively small group that we do have on our books. Yes, we'll see that sort of skill base decline as we move the entire business to technology solutions. And as Tony said, we'll still have these services or human wrapper. It won't disappear entirely in the near future. But it will run over progressively. So on the cost of sales, we have that under control. And as you can see from the gross margin, we continue to actually do that very well because we're actually seeing albeit a slight increase in some periods, but we're seeing it consistent in cost of sales. In the OpEx, yes, there are some roles that we will refocus. They might have been coordinating multiple services contracts. We'll have more and more technology contracts. Some of those people will shift, but there may be some decline in those people costs. I think other costs, which is sort of the thing that we will be addressing, Tony mentioned Dark Fiber. All of those sorts of costs disappear. So that could be -- in Australia, it could be several hundred thousand dollars of those sorts of costs that will disappear progressively. We know how to -- we've been reducing lots of other costs. The human cost, I pointed out when I looked at the results is the one area that's increased, but that's just the nature of engineering and sales and others. So we do expect some cost containment, and we are very, very focused on doing that in a timely fashion. So I think the real outcome is that we won't see a negative impact as we lose the services revenue or we convert the services revenue to technical type revenue.

We've just got our last 2 questions. So while we're talking about financials, JB, what is the outlook for OpEx growth? Will investment in sales and marketing continue to drive 10-plus growth? Or is it reasonable to expect to see some operating leverage coming through in addition to GM expansion?

We expect some operating leverage to come through. You're growing a business. We're growing into new territories. We're building relationships with value-added resellers and similar. Yes, we do expect those costs to see some increase, but we don't expect it to be a serious -- we're not in the forecast. It will grow, but it will grow in coordination with the business growing, but we should see the improvement in the gross margin should accelerate beyond the cost.

So just down to the last 2 questions. So first one, Tony, what issues surrounding the business now and upcoming keeps you up at night?

Good question. Well, it changes. When -- this time last year, it was probably where we going to find engineering talent that we need, but that's not a problem now because the market's cooled significantly. For me, I think it's -- are we attacking -- have we got sufficient sales resources to attack all of the markets that are right for LEXI right now? So in other words, are we leaving anything on the table? And we've just had a very, very good last 3 months with the Board and executive team running through exactly what that sales acceleration strategy needs to look like, whether it's a mix of indirect and direct sales opportunities. Just to flag, we're not seeing gaps in our product or technology that we need to cover by way of acquisition. I often get asked, what is an acquisition target that would be helpful? Well, if someone's got already an established base of customers that we could back our product into, then that's an obvious one, but we're not seeing lots of those opportunities out there either. Why? Because outside of broadcast, this is a fairly new technology that people are looking to implement. And so we're still at the point of having those conversations with the systems integrators, with other channel partners and with index customers who can be the lead. So it's really making sure that we're not just focused on what was yesterday's revenue, but we're focused on what tomorrow's revenue looks like. And often, it's the customers that we want to be talking to that don't even know what solution is possible for them. And that's why the only way to actually achieve that is to get in front of these people. That's why we've been investing heavily in trade shows, and that's another expense that's gone up year-on-year, and we expect that to continue to grow up. It's a very, very rich picking for us on the trade shows. I would say, well over half of our sales are actually coming in ultimately through a link or a connection to one of the major trade shows. And so what keeps me up at night is, are we spending enough? Have we got enough boots on the ground? Have we -- are we doing things the right way? Are we incentivizing the sales teams the right way? Could we be doing it better? How do we incentivize a team in a new area versus how do we incentivize them in an established area? So it's really about sales growth. What are we doing to ensure sales growth? I don't have any concerns about products. Certainly, don't have any concerns about technology. What I do have concerns about is whether we're actually unlocking all the markets that I think will get value from there, that we can lock in as quickly as possible into that iCap network so that we've got them as we roll out more and more features in the LEXI Toolkit.

Perfect. And just on our final question, regarding the revenue potential from LEXI expanding into new language captioning such as French and Spanish, what is the current volume going through iCap now in the main languages outside English? Can you give us some idea how much revenue potential there is from this expansion?

Yes. So we disclosed publicly that about 99% of traffic through iCap 6 months ago was English. With these sign-ons, we're expecting that to fall towards about 95%. But what this really shows is the capability more broadly to put in more languages, not just French and Spanish, right? Like we didn't do anything for French. We didn't do anything to make French and Spanish better. It's the underlying AI engines in French and Spanish that got better. And when they got to that point, we were able to put them up on to the iCap cloud and use all of the benefits of our encoding technology in the LEXI data interchange to make it work better than anyone had ever seen before. Now that will work just as well in Polish and Romanian and Hungarian as it will in French and Spanish, but we hadn't done the encoding work to put those language into play or those display characters back yet. So French and Spanish is -- yes, it's about French and Spanish, but it's about much more than French and Spanish. And therefore, we would ultimately see English being less than 50% of the total. But every other language basically making up the difference as we go. But we'll certainly be keeping an eye on it and highlighting certainly, which of the major languages we're getting traction on for the first time, like we are now with French and Spanish. A year ago, French and Spanish didn't work, right? And so that's why this is significant. And there are obviously very, very, very major markets in our existing territories. So for example, the first Spanish language broadcasters that we turned on were American Spanish language broadcasters, right? We were broadcasting into the U.S. Off the success of that, we then signed up the Mexican broadcasters. So that's the way we do it, right? Do it -- find a first customer, an index customer, get them to be a reference site and then rinse and repeat, but make sure we've got enough sales resources focused on delivering the right message in each of those new verticals.

And that takes us to the end of the Q&A segment. So I'll pass back to you for closing remarks.

Thanks, Mel. Thank you, everyone, for joining, and some great questions. JB and I are on a roadshow for the next month or so. So if you want a one-on-one meeting, please do reach out. We've got them in-person lunches in Sydney and Melbourne that we'd love people who are interested to come and join. And please reach out to Mel to sort out a date and a time. Obviously, welcome you also to check in with the analysts who publish material on this, and thank you for your continued support and patience. We do think this is a bit of a turning point, and we hope the market will ultimately assess us for what we see as a very strong technology solution that's bringing the best of this transformational AI technology to really important customers and markets. Starting with broadcast in the U.S. but expanding now globally in broadcast and then expanding outside of broadcast into both government and enterprise sectors and using live captioning as just the entry point for which to deliver further examples of AI language services through our LEXI Toolkit. So we're very excited for the period ahead and hope you can be part of the journey. Thanks very much, Mel.